fix(agw): Fixed SCTP abort issue by setting finite timeout in sctp_sendmsg#13146
Conversation
|
Thanks for opening a PR! 💯
Howto
More infoPlease take a moment to read through the Magma project's
If this is your first Magma PR, also consider reading
|
ssanadhya
left a comment
There was a problem hiding this comment.
Lgtm, with minor comment. Please address before merging.
| auto n = msg.size(); | ||
| auto rc = sctp_sendmsg(assoc.sd, buf, n, NULL, 0, htonl(assoc.ppid), 0, | ||
| stream, 0, 0); | ||
| stream, 100, 0); |
There was a problem hiding this comment.
Please add a comment stating that the 100 indicates a timetolive value of 100 ms.
|
@bhuvaneshne , please run LTE integ tests on this change and update the test plan. |
|
This needs to be rebased on master, as there are foreign commits in here, which should not be on this PR. |
Neudrino
left a comment
There was a problem hiding this comment.
I am removing all the reviewers, as they are only on here due to bad commits in the branch!
Hi @ssanadhya All the cleanups happen in the scenario a test case fails including cleaning the interfaces and stopping the unclosed iperf instances. If any test case fails, including data test case, we need not reload the VMs. Hi @bhuvaneshne Although route should not be an issue here for uplink data test cases, can you please add route in the TRF Server VM with below command before running the test case? Now if you still have issues with local setup, since the branch has already been created and raised for PR, there is one more quick approach for you to verify the integ test directly via Github Actions:
|
374d118 to
2109ac4
Compare
Thanks @VinashakAnkitAman , I rebased and reran the test on a new environment. It failed again but in a different spot. Let me try the "run workflow" |
2109ac4 to
8f1ea21
Compare
@bhuvaneshne By any chance you got the integ test working over this PR?
@bhuvaneshne By any chance you got the integ test working over this PR? |
Hi @VinashakAnkitAman : the test reported failure. I did not retry once again - I will give it a shot once again and update. |
|
@bhuvaneshne , it will be great to have this fix in the release 1.8 (which was branched out last week). If you could please confirm that LTE integration tests are passing on this PR, we can merge and backport this change to the official release. @VinashakAnkitAman , can you also test the changes of this PR in your local setup? |
Sure @ssanadhya I am on leave on Tuesday for some sudden personal work.. I will re-verify in my local setup as well on Wednesday morning.. |
Hi @ssanadhya , Below are the test cases that fail even after retries: I can rebase and give it one more try |
8f1ea21 to
e6c6103
Compare
|
@Neudrino , could you please review this revised PR? It is awaiting your approval since you requested changes. |
|
@ssanadhya , Finally got a clean run (Had to retry couple of test cases). Attaching the test results |
…g resolved. magma#13115 Signed-off-by: bhuvaneshne <[email protected]>
Signed-off-by: bhuvaneshne <[email protected]>
e6c6103 to
bdc312c
Compare
|
Thanks for confirming @bhuvaneshne ! I will merge this once CI checks are complete.
|
The change request is no longer required, as the appropriate parties reviewed.
…ndmsg (#13146) * fix(agw): SCTP Abort on new connections from ENB. MME stops processing resolved. #13115 Signed-off-by: bhuvaneshne <[email protected]> (cherry picked from commit 6506c85)
💚 All backports created successfully
Questions ?Please refer to the Backport tool documentation and see the Github Action logs for details |
…ndmsg (#13146) (#13644) * fix(agw): SCTP Abort on new connections from ENB. MME stops processing resolved. #13115 Signed-off-by: bhuvaneshne <[email protected]> (cherry picked from commit 6506c85) Co-authored-by: Bhuvanesh <[email protected]>
|
already in 1.8 - setting respective label |

Signed-off-by: bhuvaneshne [email protected]
Summary
See bug description (#13115) for description, RCA and proposal of fix.
Test Plan
Reproduce the issue using the bundled simulator (See: #13115)
Replace the sctpd executable
Restart sctp daemon (systemctl restart sctpd)
Wait for sctpd to get ready (See syslog)
Rerun the simulator and see that the issue is not reproducible again
Additional Information